A genome-wide RNAi screen identifies regulators of cholesterol-modified hedgehog secretion in Drosophila

Abstract

Hedgehog (Hh) proteins are secreted molecules that function as organizers in animal development. In addition to being palmitoylated, Hh is the only metazoan protein known to possess a covalently-linked cholesterol moiety. The absence of either modification severely disrupts the organization of numerous tissues during development. It is currently not known how lipid-modified Hh is secreted and released from producing cells. We have performed a genome-wide RNAi screen in Drosophila melanogaster cells to identify regulators of Hh secretion. We found that cholesterol-modified Hh secretion is strongly dependent on coat protein complex I (COPI) but not COPII vesicles, suggesting that cholesterol modification alters the movement of Hh through the early secretory pathway. We provide evidence that both proteolysis and cholesterol modification are necessary for the efficient trafficking of Hh through the ER and Golgi. Finally, we identified several putative regulators of protein secretion and demonstrate a role for some of these genes in Hh and Wingless (Wg) morphogen secretion in vivo. These data open new perspectives for studying how morphogen secretion is regulated, as well as provide insight into regulation of lipid-modified protein secretion.

Figure 1. Generation of a secreted, biologically active Hh-Renilla fusion protein.

(A) Schematic representation of the secreted Hh-Ren constructs. The coding sequence for Renilla luciferase was inserted upstream of the autoproteolytic cleavage site of Hh such that the processed N-terminal protein would be fused to Renilla. Hh-Ren fusions lacking either the cholesterol (HhN-Ren) or both cholesterol and palmitate modifications (HhC85SN-Ren) were also generated. (B) Western blot of cell lysates from S2 cells expressing different Renilla fusions constructs showing the full-length (FL) and N-terminal proteins (N). (C) Processing of Hh-Ren is increased by addition of cholesterol. S2 cells expressing Hh-Ren were cultured in the presence or absence of a cholesterol lipid concentrate. After 5 days, the cells were lysed and examined by western blot analysis using a Hh-specific antibody. (D) Hh-Ren activates the Hh signalling pathway. Lysates from cells expressing either wild-type Hh or Hh-Ren examined by western blot analysis for Fused (Fu) show a change in electromobility of Fu, indicative of Fu phosphorylation and pathway activation (Ctrl: untransfected S2 cells). (E) Only processed Hh-Ren is secreted. Normal S2 cells or cells expressing Hh-Ren were cultured in serum-free medium to allow detection of Hh-Ren in conditioned medium (which is normally masked by serum on a WB). After 24 h, medium was collected and cells were lysed and examined by western blot analysis using anti-Renilla. The asterisks (*) indicates a non-specific band in the conditioned medium. (F) Effect of RNAi on a stable cell line co-expressing Hh-Ren and a cytoplasmic firefly luciferase. Cells were treated with dsRNA against Ci (Ctrl), which is not expressed in S2 cells, Hh or syntaxin 5 (Syn5) for 5 days, at which point the medium was changed and cells were cultured for an additional 24 h. The bars represent the mean medium Renilla activity normalized by the lysate firefly activity ± SD.

Figure 2. A genome-wide RNAi screen for regulators of Hh secretion and release.

(A) RNAi screening procedure. Cells stably expressing Hh-Ren and a cytoplasmic firefly luciferase were treated with dsRNA for 5 days, at which point the culture medium was replaced. After 24 h, we measured the Renilla activity in the culture medium and both the Renilla and firefly activities in the cell lysates. (B–D) Scatter plots representing the duplicate z-scores for (B) medium Renilla/firefly (C) medium Renilla/lysate Renilla, and (D) lysate Renilla/firefly as the x and y coordinates for each dsRNA screened. The positive (Syn5; in red) and negative (GFP; in green) controls from each of the 58 plates are shown. Hits considered for further screening are indicated by the grey square. (E) Scatter plot representing the duplicate z-scores the firefly reading as the x and y coordinates for each dsRNA screened. The positive (thread; in red) and negative (GFP; in green) controls from each of the 58 plates are shown. Hits eliminated from further screening are indicated by the red squares. (F) Candidate regulators of Hh secretion confirmed by secondary screening. To confirm the effects of the candidate genes, new dsRNAs were synthesized which did not overlap with those used in the primary screen. S2 cells transiently transfected with an inducible pMT-Hh-Ren construct and a cytoplasmic firefly luciferase were cultured for 5 days with the indicated dsRNAs, at which point the medium was replaced with Cu-containing medium, and the cells were cultured for an additional 36 h. The bars represent the mean medium Renilla activity normalized by the lysate firefly activity ± SD. Genes identified in previous screens for regulators of general secretion in Drosophila cells , are indicated in red. Ctrl: dsRNA against Ci.

Figure 3. Cleavage and cholesterol modification affect the subcellular trafficking of Hh.

(A–C) S2 cells were transiently transfected with pMT-Hh-Ren (A,B), pMT-HhN-Ren, or pMT-HhC85SN-Ren (C) and cultured for 5 days with the indicated dsRNAs and analyzed as in Figure 2F. The bars represent the mean medium/lysate Renilla activity ± SD. Similar results were obtained using several independent dsRNAs (Table S4). Ctrl: dsRNA against Ci. (D) COPI, but not COPII, is required for Hh secretion. S2R+ cells transiently transfected with an inducible pMT-Hh construct were treated with the indicated dsRNAs and cultured for 5 days, at which point the medium was replaced with Cu-containing medium, and the cells were cultured for an additional 36 h. Medium was collected and submitted to high speed spin and cells were lysed and examined by western blot analysis using anti-Hh. Similar results were obtained with two independent amplicons targeting Sec23: Sec23-1 (DRSC31248) and Sec23-2 (DRSC12387). (E) Lack of cholesterol modification causes Hh accumulation in the ER and Golgi. S2 cells were transfected with Hh, fixed and immunostained with anti-Hh (green) and anti-Golgi (red). In contrast to Hh, HhN was strongly localized to the Golgi and to the perinuclear ER (arrow). (F) Cholesterol promotes Hh secretion. S2 cells transiently transfected with an inducible pMT-Hh-Ren construct, pMT-HhN-Ren, or pMT-HhC85SN-Ren were cultured with the indicated amount of cholesterol concentrate and treated as in Figure 2F. The bars represent the mean medium/lysate Renilla activity ± SD. (G) Full length uncleaved Hh is retained in the ER. S2R+ cells were transfected with HhHA or HhUHA, fixed and immunostained with anti-HA (green) and anti-GMAP (red). HhUHA was retained in the ER, with very little co-localization with GMAP. Scale bars, 3 µm.

Figure 4. Identification of genes regulating intracellular protein traffic and Golgi structure.

(A) Putative mammalian orthologs and functional domains of candidate genes considered for further characterization. (B) S2 cells transiently transfected with pMT-Ren were cultured for 5 days with the indicated dsRNAs and treated as in Figure 2F. The bars represent the mean medium Renilla activity normalized by the lysate firefly activity ± SD. (C) S2 cells were transfected with MannII-GFP and cultured with dsRNAs against the indicated genes for 5 days. MannII-GFP staining was quantified by counting the number of cells displaying a normal, fused, or fragmented MannII-GFP pattern (see examples in Figure S5), expressed as a percentage of the total cells counted (100–200 cells per treatment). (D–I) S2 cells were cultured with dsRNAs against the indicated genes for 5 days and processed for GMAP immunostaining (red). The control cells in D were treated with Ci dsRNA. Scale bar, 3 µm. (D′–I′) S2 cells treated with dsRNA for 5 days were analyzed by EM. Note the intact Golgi stacks in D′ (arrow head) compared to the disrupted Golgi structures (arrows) or elongated fragmented Golgi remnants (between brackets) in other panels. Scale bar, 0.5 µm.

Figure 5. In vivo validation of putative regulators of protein secretion using dsRNA transgenics.

(A) Wg (green) and Hh (yellow) expression domains in the larval imaginal disc and the area patterned by Hh in the adult wing. (B) The wing phenotypes of adult flies expressing the indicated UAS-dsRNA under the control of en-Gal4. (C) Quantification of the wing intervein 3–4 domain. The intervein domain area for each wing was measured and normalized over total wing area. Results are shown as the mean percent reduction of the vein 3–4 domain relative to the control ± SD. “n” indicates the number of wings analyzed for each genotype.

Figure 6. The silencing of CG5964, CG3305, or CG12693 abolishes HhM1-induced anterior outgrowth and alters Hh distribution in producing cells.

All panels show confocal images of wing imaginal discs immunostained for Hh (green), Ptc (blue) and phalloidin (red). (A–F″) Confocal x/y sections at low (A–F) and high magnifications (A′–F″). Anterior is to the left and dorsal down. The overexpression of Hh from producing cells using hh-Gal4 generates an anterior outgrowth (arrow in A). Note the broadening of Ptc expression domain (A–A″). The expression of dsRNA against GFP (B–B″), or dsRNA against CG8441 (C–C″) together with HhM1 have no effect on HhM1-induced disc morphology and Hh signalling in receiving cells. In contrast, the overexpression of dsRNAs targeting CG5964, CG3305, or CG12693 prevents HhM1-induced anterior outgrowth (D, E, F). Note that Ptc expression is restored to normal levels in 2 rows of anterior cells (D′–F″ compared to Figure S7A–A′). (G–I′) Confocal Z sections showing Hh subcellular distribution when HhM1 is expressed alone (G–G′) or together with dsRNA against GFP (H–H′), or with dsRNA against CG8441 (I–I′). Hh is membranous and localizes mainly to the apical membrane. Note also the high accumulation of Hh at the apical plane of receiving cells. (J–L′) The silencing of CG5964 (J–J′), CG3305 (K–K′), and CG12693 (L–L′) alters Hh distribution. Hh is patchy and appears to be distributed over the entire apical/basal axis. Note also that Hh neither accumulates at the apical membrane of producing cells nor at that of receiving cells. Dashed lines mark the position of the compartment boundary (G–L) or position of the Z section within the disc (A′–F′). Apical is up and basolateral is down in panels G–L′.

Figure 7. Secretion of Wg is also impaired upon expression of dsRNA against CG5964.

(A) The wing phenotypes of adult flies expressing the indicated UAS-dsRNA under the control of hh-Gal4. (B) Immunostaining for Cut in wing imaginal discs from flies expressing the indicated UAS-dsRNA under the control of hh-Gal4. (C) Immunostaining for Wg in non-permeabilized (extracellular Wg) and permeabilized (total Wg) wing imaginal discs along the dorso-ventral axis. The broken line indicates the A/P border and the dsRNA was expressed only in the posterior compartment (to the right). The A/P border was determined by anti-Hh staining (not shown). (D) Quantification of extracellular Wg staining (see Materials and Methods for details). Bars represent the mean ratio of Wg staining intensity in the anterior divided by the Wg staining intensity of the posterior compartment ± SD for the region marked in red in C. Ctrl: dsRNA against Ci.